A question about MQA

[skbe]

As explained, if everyone is downloading the same file, it’s original +1. The same quality for everyone who downloads it.

If however one person downloads the file, and shares it with a buddy, the quality starts to deteriorate. The more often you make a copy of a copy, the higher the likelihood it stops working.


Sent from my iPad using Tapatalk

I tried a quick search on google about this and I did not find something convincing
I am not doubting your sentence but since it is very important and maybe against many people’s belief, can you share with us a serious source for this?
Thanks

 

[Kuoppis]

Basically, as an answer to both Al and skbe, have you never experienced the situation on your home computer, where you try to install a program, and it does not work? Then you delete it and install it again, and et voilà, all fine.

This is indeed very common in SW. The really intriguing thing about SW is, that you sometimes just don’t know why it does not work, even though you did everything correctly. Just happens.

Audio files are SW, as simple as that. Same with images, you transfer them and sometimes you do have artifacts for no good reason.

I think Mike L. summarized it in a nice way, there are a lots of variables in every sharing event.


Sent from my iPad using Tapatalk

 

[Kuoppis]

Just for the benefit of everyone, this is not something I just came up with. The terminus technicus is files do get “corrupted”.

I’m sure there’s plenty to read about that on the interweb.


Sent from my iPad using Tapatalk

 

[NekoAudio]

As Al said, PCM is better described as a lossless encoding scheme, and not as a compression scheme, because the original data values are represented by individual codes, and there is no reduction of multiple values to a fewer number of codes.

MQA is indeed lossy, because the data output after decode (either with our without MQA decode) is not identical to the original input data.

Digital files transferred or copied using a reliable protocol will not degrade in quality. Comparing [software installation bugs or code incompatibilities that result in a program not working] or [image transfer using an application or service that applies its own compression to save time or money] or [data corruption / bit-rot] to [general purpose reliable data transfer] is fallacious.

In the context of different audio encodings or distributors, provenance is usually taken to mean knowledge of the "chain of custody" and whatever happened along the way.

 

[Kuoppis]

As Al said, PCM is better described as a lossless encoding scheme, and not as a compression scheme, because the original data values are represented by individual codes, and there is no reduction of multiple values to a fewer number of codes.

MQA is indeed lossy, because the data output after decode (either with our without MQA decode) is not identical to the original input data.

Digital files transferred or copied using a reliable protocol will not degrade in quality. Comparing [software installation bugs or code incompatibilities that result in a program not working] or [image transfer using an application or service that applies its own compression to save time or money] or [data corruption / bit-rot] to [general purpose reliable data transfer] is fallacious.

In the context of different audio encodings or distributors, provenance is usually taken to mean knowledge of the "chain of custody" and whatever happened along the way.

What does encoding/ decoding mean technically? It means packing and unpacking, which essentially is compression.

I.e. making the information package smaller and then bigger again.


Sent from my iPad using Tapatalk

 

[NekoAudio]

What does encoding/ decoding mean technically? It means packing and unpacking, which essentially is compression.

I.e. making the information package smaller and then bigger again.

In the realm of computer science, the term compression is not used if simply talking about making the representation smaller. For example, in the following example of morse code for the letter 'C', the second encoding would not commonly be called compression:

1. dash dot dash dot
2. -.-.

Even though the second line uses 4 bytes of data, compared to the first line using 17 bytes, this would be called encoding and not compression.

Encoding means to convert one representation into another. Decoding the reverse operation.

 

[Kuoppis]

MQA is indeed lossy, because the data output after decode (either with our without MQA decode) is not identical to the original input data.[/URL].

Slightly incorrect, I’m afraid.

a) yes, it is lossy if not rendered and decoded, or only rendered (i.e. unpacked with SW only and w/o the second level MQA decoder). They call that first level unfolding.

b) no, it is not lossy if rendered and decoded with a MQA decoder. However, the information is not identical, as parts from the inaudible band are transported in the message footer to allow for a smaller file size. That does not mean the same information is not there, it’s just in another place.

But this is nothing new, can be read everywhere. There is just this statement from Andreas Koch people are gettin hung up about. Of course he is against MQA, as they are his competitor and eating his lunch.

Andreas Koch just spent a decade arguing why DSD is good, and faced exactly the same kind of pushback as MQA does. Sony was not much better than Meridian/ MQA trying to achieve world domination with SACD. At first Linn was very vocal about DSD being nonsensical, but now they support it. Now they argue MQA does not make sense. See a pattern here?


Sent from my iPad using Tapatalk

 

[Kuoppis]

In the realm of computer science, the term compression is not used if simply talking about making the representation smaller. For example, in the following example of morse code for the letter 'C', the second encoding would not commonly be called compression:

1. dash dot dash dot
2. -.-.

Even though the second line uses 4 bytes of data, compared to the first line using 17 bytes, this would be called encoding and not compression.

Encoding means to convert one representation into another. Decoding the reverse operation.

I do agree, your morse example has nothing to do with compression.

If the encoding makes the data package smaller and decoding bigger again, it is a form of compression.

Sent from my iPad using Tapatalk

 

[mep]

Just for the benefit of everyone, this is not something I just came up with. The terminus technicus is files do get “corrupted”.

I’m sure there’s plenty to read about that on the interweb.


Sent from my iPad using Tapatalk

If you didn’t make this up, you should really point us to where we can read the source of your belief that “The quality of source file deteriorates similarly to an analog master when you make a copy of a copy of a copy etc. Deterioration is just slower.“

You won’t find it though unless you read it on The Onion.

 

[skbe]

If you didn’t make this up, you should really point us to where we can read the source of your belief that “The quality of source file deteriorates similarly to an analog master when you make a copy of a copy of a copy etc. Deterioration is just slower.“

You won’t find it though unless you read it on The Onion.

I agree
And please not an article in latin nor in greek
Because talkum in latinum does not meanus not bullshitum
Et voilà, ceci veut dire cela
Au revoir

 

[NekoAudio]

b) no, it is not lossy if rendered and decoded with a MQA decoder.

A 24-bit/96kHz data MQA decode is not going to equal the original 24-bit/96kHz data that was input to the MQA encoder. From the patent application.

I do agree, your morse example has nothing to do with compression.

If the encoding makes the data package smaller and decoding bigger again, it is a form of compression.

In the morse code example, the data package is indeed smaller after encoding (translation from written words to abbreviated . and - characters). It can be referred to as compression, but typically isn't.

 

[Kuoppis]

A 24-bit/96kHz data MQA decode is not going to equal the original 24-bit/96kHz data that was input to the MQA encoder. From the patent application.

Exactly. As explained earlier, data from the inaudible spectrum is moved to the footer. Of course the file is different, that’s the whole point. Why is that difficult to understand?

In the morse code example, the data package is indeed smaller after encoding (translation from written words to abbreviated . and - characters). It can be referred to as compression, but typically isn't.

Compression here is referring to an audio signal, i.e. compression within the same medium. Again, totally different topic.


Sent from my iPad using Tapatalk

 

[Kuoppis]

If you didn’t make this up, you should really point us to where we can read the source of your belief that “The quality of source file deteriorates similarly to an analog master when you make a copy of a copy of a copy etc. Deterioration is just slower.“

You won’t find it though unless you read it on The Onion.

Start here, it’s in color. There might be an audiobook version in case reading is troublesome.

Sent from my iPad using Tapatalk

 

[Kuoppis]

Once the music is on your NAS, is it safe and secure forever? Maybe not.

From Wikipedia:
Data corruption refers to errors in computer data that occur during writing, reading, storage, transmission, or processing, which introduce unintended changes to the original data. Computer, transmission, and storage systems use a number of measures to provide end-to-end data integrity, or lack of errors.
https://en.wikipedia.org/wiki/Data_corruption

From Sage:
Damaged files or corrupt data of some variety on a computer is inevitable. It's not a matter of if, it's a matter of when. Computers are by and large, extremely reliable machines, but they are still machines and are subject to wear and tear just like anything else.
https://support.na.sage.com/selfservice/viewContent.do?externalId=69520&sliceId=1

Practical test case from CERN:
https://www.zdnet.com/article/data-corruption-is-worse-than-you-know/


Sent from my iPad using Tapatalk

 

[NekoAudio]

Exactly. As explained earlier, data from the inaudible spectrum is moved to the footer. Of course the file is different, that’s the whole point. Why is that difficult to understand?

I'm not referring to the file, or to the encoded data being different than the pre-encoded data. I'm referring to whether or not the audio data pre-encode and post-decode is identical or not. If you input a sequence of 24-bit/96kHz audio data into the MQA encode process for encoding into 24-bit/96kHz data, the audio data that comes out of the MQA decode will not be identical to the original input. For starters, input data would be truncated to 17-bits as shown in Figure 7A.

Compression here is referring to an audio signal, i.e. compression within the same medium. Again, totally different topic.

I think it is okay if you want to use compression only in the sense of the bits necessary for a single code being fewer than when using a different code. I was just pointing out that within the computer science and engineering community we generally wouldn't do so, and would more likely find that usage confusing. The "medium" is irrelevant ("audio signal" isn't a medium)—data compression is area of science and technology that should only be audio-specific when the compression algorithm is lossy and wants to incorporate psychoacoustic research to minimize perceived quality loss.

 

[Kuoppis]

I'm not referring to the file, or to the encoded data being different than the pre-encoded data. I'm referring to whether or not the audio data pre-encode and post-decode is identical or not. If you input a sequence of 24-bit/96kHz audio data into the MQA encode process for encoding into 24-bit/96kHz data, the audio data that comes out of the MQA decode will not be identical to the original input. For starters, input data would be truncated to 17-bits as shown in Figure 7A.


I think it is okay if you want to use compression only in the sense of the bits necessary for a single code being fewer than when using a different code. I was just pointing out that within the computer science and engineering community we generally wouldn't do so, and would more likely find that usage confusing. The "medium" is irrelevant ("audio signal" isn't a medium)—data compression is area of science and technology that should only be audio-specific when the compression algorithm is lossy and wants to incorporate psychoacoustic research to minimize perceived quality loss.

We’ve been through this, already twenty posts ago.

All data is still there, it’s just in a different place. And it can’t be fully decoded w/o a MQA renderer/ decoder.

 

[mep]

Start here, it’s in color. There might be an audiobook version in case reading is troublesome.

Sent from my iPad using Tapatalk

This all you have? If there was ANYTHING in that book that backed up your ridiculous statement that copying a digital file causes degradation to the file, you would have pulled it out of the book and bolded it. You have nothing.

 

[mep]

Once the music is on your NAS, is it safe and secure forever? Maybe not.

From Wikipedia:
Data corruption refers to errors in computer data that occur during writing, reading, storage, transmission, or processing, which introduce unintended changes to the original data. Computer, transmission, and storage systems use a number of measures to provide end-to-end data integrity, or lack of errors.
https://en.wikipedia.org/wiki/Data_corruption

From Sage:
Damaged files or corrupt data of some variety on a computer is inevitable. It's not a matter of if, it's a matter of when. Computers are by and large, extremely reliable machines, but they are still machines and are subject to wear and tear just like anything else.
https://support.na.sage.com/selfservice/viewContent.do?externalId=69520&sliceId=1

Practical test case from CERN:
https://www.zdnet.com/article/data-corruption-is-worse-than-you-know/


Sent from my iPad using Tapatalk

Dear mep, you know better thank trying to convince me with a Wikipedia quote.

So why are you using a Wikipedia quote?

 

[rbbert]

From the patent application, MQA is "essentially" lossless only at half the sampling rate of the original file; i.e., 24/192 can be "essentially lossless" only in a 24/96 MQA container, and a 24/48 container can only contain "essentially lossless" information up to 24/96. At sampling rates more than twice the original, MQA is acknowledged in its own patent application as lossy. In it's "essentially lossless" guise, MQA files are about the same size as corresponding FLAC files, so it doesn't really provide a size advantage either.

 

[steve59]

From the patent application, MQA is "essentially" lossless only at half the sampling rate of the original file; i.e., 24/192 can be "essentially lossless" only in a 24/96 MQA container, and a 24/48 container can only contain "essentially lossless" information up to 24/96. At sampling rates more than twice the original, MQA is acknowledged in its own patent application as lossy. In it's "essentially lossless" guise, MQA files are about the same size as corresponding FLAC files, so it doesn't really provide a size advantage either.

If the format is sending the information at 1/2 of 1/3 the file size how is it the same size? I'm reading your post, but not interpeting.